Add goal message to Fate

[milleruntime]

• Closes Allow FATE operations to report goal and status information. #1249

[milleruntime]

Here are some examples of log output:

2022-02-14T10:25:34,906 [fate.Fate] INFO : FATE TxId: 3799578337887111047 Op: TABLE_CREATE Create table ci ONLINE with 19 splits.
2022-02-14T10:37:48,674 [fate.Fate] INFO : FATE TxId: 5505785720937952578 Op: TABLE_COMPACT Compact table (1) with config []
2022-02-14T10:38:18,256 [fate.Fate] INFO : FATE TxId: 2554407174635375739 Op: TABLE_DELETE Delete table ci(1)

[ctubbsii]

@milleruntime I think GitHub is supposed to auto-link PRs to their referenced issues. However, I don't think it works the way you're doing it. It should work when you add the closing keywords to the description or body of a commit message, but it does not seem to work when you add it to the subject line. For consistency (and for less confusion), it's probably best to put these in the body of the commit anyway, so that way when GitHub appends the PR number on merge, it looks a bit nicer:

Add goal message to Fate (#2492)

vs.

Add goal message to Fate. Closes #1249 (#2492)

[EdColeman]

From the sample log

2022-02-14T10:37:48,674 [fate.Fate] INFO : FATE TxId: 5505785720937952578 Op: TABLE_COMPACT Compact table (1) with config []
2022-02-14T10:38:18,256 [fate.Fate] INFO : FATE TxId: 2554407174635375739 Op: TABLE_DELETE Delete table ci(1)

What is the number (1) in parentheses? The tableId?

[milleruntime]

From the sample log

2022-02-14T10:37:48,674 [fate.Fate] INFO : FATE TxId: 5505785720937952578 Op: TABLE_COMPACT Compact table (1) with config []
2022-02-14T10:38:18,256 [fate.Fate] INFO : FATE TxId: 2554407174635375739 Op: TABLE_DELETE Delete table ci(1)

What is the number (1) in parentheses? The tableId?

Yes

[milleruntime]

This method was created because the code used to log fate txids in many different ways (hex and decimal for example)

I forgot about that, thanks @keith-turner. What do you think of changing the transaction ID in the next version to be something more straight forward? Maybe use something like java.util.UUID?

[keith-turner]

What is the number (1) in parentheses? The tableId?

In another comment on this thread I suggested a specific way to log FATE IDs that was consistent. The comments about table ids made me think it would be nice to consitently log all ids in the logs. Maybe something like <type>[<id>]. For example FATE[123], TABLE_ID[456], ECID[abd], NAMESPACE_ID[xyz]. I think I will open an issue for this.

What do you think of changing the transaction ID in the next version to be something more straight forward? Maybe use something like java.util.UUID?

@milleruntime I don't have an opinion. What is the motivation for the change?

[milleruntime]

What do you think of changing the transaction ID in the next version to be something more straight forward? Maybe use something like java.util.UUID?

@milleruntime I don't have an opinion. What is the motivation for the change?

I was just thinking that a stronger type could prevent coding errors like the one I made forgetting to call the method to display the value in HEX.

[keith-turner]

I was just thinking that a stronger type could prevent coding errors like the one I made forgetting to call the method to display the value in HEX.

Oh yeah, the code would probably be 10x better if we had a FateTxId type that we used instead of long. Its toString method could consistently format for logging maybe. That could possible encapsulate or extend UUID, but I think we would benefit form a more specific type than UUID or long.

[ctubbsii]

I was just thinking that a stronger type could prevent coding errors like the one I made forgetting to call the method to display the value in HEX.

Oh yeah, the code would probably be 10x better if we had a FateTxId type that we used instead of long. Its toString method could consistently format for logging maybe. That could possible encapsulate or extend UUID, but I think we would benefit form a more specific type than UUID or long.

That seems like a bigger change than updating logging. Should that be considered as part of #2473 ?

[keith-turner]

That seems like a bigger change than updating logging. Should that be considered as part of #2473 ?

I was not thinking that change should be made on this PR. Just seems like a good general change to make. I think introducing a fate tx id type to replace long would be a nice change to make by itself for the purposes of review.